Alcohol and nicotine polygenic scores are associated with the development of alcohol and nicotine use problems from adolescence to young adulthood

Background and Aims Molecular genetic studies of alcohol and nicotine use have identified many genome-wide association study (GWAS) loci. We measured associations between drinking and smoking polygenic scores (PGS) and trajectories of alcohol and nicotine use outcomes from late childhood to early adulthood, substance-specific versus broader-liability PGS effects, and if PGS performance varied for consumption versus problematic substance use. Design, setting, participants and measurements We fitted latent growth curve models with structured residuals to scores on measures of alcohol and nicotine use and problems from ages 14 to 34 years. We then estimated associations between the intercept (initial status) and slope (rate of change) parameters and PGSs for drinks per week (DPW), problematic alcohol use (PAU), cigarettes per day (CPD) and ever being a regular smoker (SMK), controlling for sex and genetic principal components. All data were analyzed in the United States. PGSs were calculated for participants of the Minnesota Twin Family Study (n = 3225) using results from the largest GWAS of alcohol and nicotine consumption and problematic use to date. Findings Each PGS was associated with trajectories of use for their respective substances [i.e. DPW (beta(mean) = 0.08; beta(range) = 0.02-0.12) and PAU (beta(mean) = 0.12; beta(range) = -0.02 to -0.31) for alcohol; CPD (beta(mean) = 0.08; beta(range) = 0.04-0.14) and SMK (beta(mean) = 0.18; beta(range) = 0.05-0.36) for nicotine]. The PAU and SMK PGSs also exhibited cross-substance associations (i.e. PAU for nicotine-specific intercepts and SMK for alcohol intercepts and slope). All identified SMK PGS effects remained as significant predictors of nicotine and alcohol trajectories (beta(mean) = 0.15; beta(range) = 0.02-0.33), even after adjusting for the respective effects of all other PGSs. Conclusions Substance use-related polygenic scores (PGSs) vary in the strength and generality versus specificity of their associations with substance use and problems over time. The regular smoking PGS appears to be a robust predictor of substance use trajectories and seems to measure both nicotine-specific and non-specific genetic liability for substance use, and potentially externalizing problems in general.